What we attend to at any moment determines what we learn at that moment, and this also depends on our past learning. This grant investigates how words, perceptual cues, and category organizations form a system of associations that contextually cue attention, making it exquisitely tuned to context and thus constraining and propelling future learning. Although old views characterized associative learning as simple counting of individual pairings, more advanced theories suggest that people learn systems of overlapping associations that (1) lead to systematic effects on attention, (2) tune learning to the context, and (3) give rise to higher-order (almost rule-like) correlations. This grant seeks a mechanistic understanding of these processes in word learning context. Twenty- four to 48-months-old children will participate in the experiments containing artificial language learning, training tasks, and the comparison of children learning two different languages (English and Japanese), with different systems of correlations. Understanding the basic processes through which context guides attention and how and why language might be a particularly potent cue to attention is a first step toward understanding the cascading consequences of language delay on cognitive development and the invention of new methods for tuning attention when language is not easily available to the learner.